mlsanityData to 14 Oct 2026

Sonata: A Hybrid World Model for Inertial Kinematics under Clinical Data Scarcity

arXiv 2604.18058 ↗First posted 20 Apr 2026cs.LGcited by 2

On the map: World models → Medical and surgical world models

Abstract (arXiv)

We introduce Sonata, a compact latent world model for six-axis trunk IMU representation learning under clinical data scarcity. Clinical cohorts typically comprise tens to hundreds of patients, making web-scale masked-reconstruction objectives poorly matched to the problem. Sonata is a 3.77 M-parameter hybrid model, pre-trained on a harmonised corpus of nine public datasets (739 subjects, 190k windows) with a latent world-model objective that predicts future state rather than reconstructing raw sensor traces. In a controlled comparison against a matched autoregressive forecasting baseline (MAE) on the same backbone, Sonata yields consistently stronger frozen-probe clinical discrimination, prospective fall-risk prediction, and cross-cohort transfer across a 14-arm evaluation suite, while producing higher-rank, more structured latent representations. At 3.77 M parameters the model is compatible with on-device wearable inference, offering a step toward general kinematic world models for neurological assessment.

Key facts. Cited by 2 papers in the map (data through 14 Oct 2026). Placed in Medical and surgical world models, one of 11 directions in World models. Builds on 34 of its 60 references that are in the map. First posted on arXiv 20 Apr 2026.

Builds on

34 of 60 references.

Closest ancestors · 1

Same direction, newest first

  1. 2025Beyond Generative AI: World Models for Clinical Prediction, Counterfactuals, and Planning20 Nov

Other references · 33

  1. 2024Transformers are SSMs: Generalized Models and Efficient Algorithms Through Structured State Space Dualitycited by 1,336
  2. 2024Chronos: Learning the Language of Time Seriescited by 759
  3. 2023A decoder-only foundation model for time-series forecastingcited by 692
  4. 2024Gated Delta Networks: Improving Mamba2 with Delta Rulecited by 532
  5. 2021Linear Transformers Are Secretly Fast Weight Programmerscited by 502
  6. 2024Parallelizing Linear Transformers with the Delta Rule over Sequence Lengthcited by 396
  7. 2026LeWorldModel: Stable End-to-End Joint-Embedding Predictive Architecture from Pixelscited by 273
  8. 2025LeJEPA: Provable and Scalable Self-Supervised Learning Without the Heuristicscited by 254
  9. 2025Kimi Linear: An Expressive, Efficient Attention Architecturecited by 196
  10. 2017Data Augmentation of Wearable Sensor Data for Parkinson's Disease Monitoring using Convolutional Neural Networkscited by 131
  11. 2020An Empirical Survey of Data Augmentation for Time Series Classification with Neural Networkscited by 128
  12. 2024Unlocking State-Tracking in Linear RNNs Through Negative Eigenvaluescited by 101
21 more
  1. 2024Moirai-MoE: Empowering Time Series Foundation Models with Sparse Mixture of Expertscited by 95
  2. 2024Timer-XL: Long-Context Transformers for Unified Time Series Forecastingcited by 87
  3. 2025DeltaProduct: Improving State-Tracking in Linear RNNs via Householder Productscited by 75
  4. 2025This Time is Different: An Observability Perspective on Time Series Foundation Modelscited by 68
  5. 2024Scaling Wearable Foundation Modelscited by 63
  6. 2020Contrastive Predictive Coding for Human Activity Recognitioncited by 55
  7. 2025SensorLM: Learning the Language of Wearable Sensorscited by 49
  8. 2025A Systematic Analysis of Hybrid Linear Attentioncited by 39
  9. 2024Toward Foundation Model for Multivariate Wearable Sensing of Physiological Signalscited by 32
  10. 2026Olmo Hybrid: From Theory to Practice and Backcited by 22
  11. 2024RelCon: Relative Contrastive Learning for a Motion Foundation Model for Wearable Datacited by 20
  12. 2025HiMAE: Hierarchical Masked Autoencoders Discover Resolution-Specific Structure in Wearable Time Seriescited by 19
  13. 2026Hybrid Linear Attention Done Right: Efficient Distillation and Effective Architectures for Extremely Long Contextscited by 18
  14. 2024Wearable Accelerometer Foundation Models for Health via Knowledge Distillationcited by 16
  15. 2026EchoJEPA: A Latent Predictive Foundation Model for Echocardiographycited by 13
  16. 2026Reverso: Efficient Time Series Foundation Models for Zero-shot Forecastingcited by 13
  17. 2024PRIMUS: Pretraining IMU Encoders with Multimodal Self-Supervisioncited by 10
  18. 2025medDreamer: Model-Based Reinforcement Learning with Latent Imagination on Complex EHRs for Clinical Decision Supportcited by 9
  19. 2025Exact Flow Linear Attention: Exact Solution from Continuous-Time Dynamicscited by 4
  20. 2026Laya: A LeJEPA Approach to EEG via Latent Prediction over Reconstructioncited by 4
  21. 2025Self-Supervised Dynamical System Representations for Physiological Time-Seriescited by 3

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 2.

2 papers.

  1. 2026MorphologyFM: A Foundation Model for Morphology-Aware Representation Learning from ECG and Pulse Oximetry Waveformscited by 0
  2. 2026Capturing Cardiac Cyclicity through Phase-Equivariant Self-Supervised Learningcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025KinTwin: Imitation Learning with Torque and Muscle Driven Biomechanical Models Enables Precise Replication of Able-Bodied and Impaired Movement from Markerless Motion Capture
  2. 2025Wearable-informed generative digital avatars predict task-conditioned post-stroke locomotion
  3. 2026CalTwin: Towards Calibrated, Shift-Robust Medical World Models via Fisher-Information Regularisation
  4. 2026ChronoMedicalWorld: A Medical World Model for Learning Patient Trajectories from Longitudinal Care Data
  5. 2026Skeletal Prototypes on Iterative Nerve Expansions
  6. 2026EquiFusion: Kinematics-Agnostic Human Motion Prediction via Equivariant Latent Diffusion
  7. 2025Refining Diffusion Models for Motion Synthesis with an Acceleration Loss to Generate Realistic IMU Data
  8. 2026Medical world models in healthcare: foundations, applications, and challenges for trustworthy clinical translation